Congenital hypothyroidism (CH) is one of the most common preventable causes of intellectual disability, and newborn screening programs worldwide have greatly improved the early detection and treatment of this condition [1]. The incidence of CH in term infants is estimated to be approximately 1 in 2,000– 3,000 live births [2]. In preterm infants, especially in very low birth weight (VLBW), thyroid dysfunction is more common due to the immaturity of the hypothalamic-pituitary-thyroid (HPT) axis, postnatal illness, iodine imbalance and medications affecting thyroid function [3]. The reported incidence of CH in VLBW infants varies across studies, ranging from 1 in 250 to 1 in 60, and is substantially higher than that in term infants [4,5].

Thyroid dysfunction in VLBW infants often manifests as delayed thyroid-stimulating hormone (TSH) elevation, where initial screening results may appear normal but TSH levels rise significantly several weeks later. Due to this risk, many VLBW infants undergo repeat thyroid screenings and close monitoring during the neonatal period. Recent guidelines recommended routine re-screening of all preterm infants for CH to catch delayed TSH elevation [6].

Currently, most clinicians treat all infants diagnosed with CH with L-thyroxine (LT4) replacement therapy and reassess the need for continued treatment at around 3 years of age to distinguish transient CH (TCH) from permanent CH (PCH) [7]. While prolonged treatment is crucial for preventing neurodevelopmental impairment (NDI), it also results in some infants who receive unnecessary medication for years. If reliable predictors for distinguishing PCH from TCH could be identified, they could help reduce unnecessary treatment duration. Proposed predictors for PCH include extreme prematurity, maternal history of thyroid disease, the degree of TSH elevation, and the LT4 dose requirement to maintain normal thyroid levels, although no consensus on early markers has been established [8].

Understanding the proportion of PCH and TCH in this population is essential for improving clinical decision-making and minimizing unnecessary long-term treatment. Therefore, this study aims to determine the incidence of PCH in Korean VLBW infants, compare it with previously reported rates, and identify clinical characteristics that may aid in earlier differentiation between TCH and PCH.

A total of 73 VLBW infants were included in the final analysis (Figure 1). Among them, 50 infants (68.5%) were classified as TCH, and 23 (31.5%) as having PCH. The mean gestational age of the cohort was 28.4±3.32 weeks, and the mean birth weight was 840±310 g. A total of 87.6% of the infants received antenatal steroids and 39.7% were treated with postnatal dexamethasone for the management of BPD. Major congenital anomalies were identified in nine infants. These included one case of trisomy 21 with atrioventricular septal defect (AVSD), three cases of VACTERL (vertebrae, anus, cardiac, tracheoesophageal fistula, renal, limb) association, two cases of ventricular septal defect, two cases of congenital diaphragmatic hernia, and one case of omphalocele. All anomalies were non-overlapping except for the trisomy 21 case with AVSD. Maternal thyroid disease was reported in 12.3% of cases.

Regarding thyroid-related parameters, the mean screening TSH level was 6.39±15.72 μIU/mL. The mean serum TSH and free T4 level before initiation of LT4 supplementation was 47.8±25.4 μIU/mL and 0.76±0.46 ng/dL, respectively. LT4 was initiated at a mean postnatal age of 37.7±20.2 days. During the follow-up period after NICU discharge, the mean LT4 requirement was 3.03±1.09 and 2.27±0.94 μg/kg/day at 12 and 36 months, respectively. Thyroid ultrasonography revealed a eutopic gland in all patients; among the seven with abnormal findings, six had a small thyroid and one had an enlarged gland.

We compared clinical characteristics, thyroid hormonerelated factors and developmental outcomes between the TCH and PCH group (Table 1). There was a trend toward higher screening TSH levels in the PCH group compared to the TCH group (6.5±18.8 vs. 6.1±4.3, P=0.054) (Figure 2A). In univariate analysis, major anomaly, the use of dexamethasone for the management of BPD and serum TSH >10 μIU/mL were significantly associated with PCH (Table 2). During the follow-up period, LT4 requirement at both 12 and 36 months was significantly higher in the PCH group compared to the TCH group (Figure 2B). In multivariate analysis, gestational age, the presence of major congenital anomalies, screening TSH >10 μIU/mL, and higher LT4 requirement dose at 36 months were significantly associated with PCH in VLBW infants with CH.

In this study, the overall incidence of CH in VLBW infants was 9.4%, with PCH accounting for 1.8%, both of which are higher than previous studies. Gestational age, the presence of major congenital anomalies, screening TSH >10 μIU/mL, and higher LT4 requirement dose at 36 months were significantly associated with PCH in VLBW infants.

The previously reported incidence of CH in term newborns ranges from 1:2,000 to 4,000 live births, with some studies suggesting rates as high as 1:1,714 depending on screening strategies or population characteristics [11-14]. The incidence of CH is substantially higher in preterm infants. A study from New England reported an incidence of 1:250 among VLBW infants [4]. Little is known about the incidence of CH, including PCH in Korea. To our knowledge, two Korean studies have investigated the incidence of CH in preterm infants. One study reported a 19.4% LT4 treatment rate in infants born before 32 weeks of gestation, while the other study found an overall CH incidence of 1.7% (approximately 1:60) in preterm infants, with an increased risk of CH risk with decreasing gestational age [5,15]. A study by Jo et al. [5] reported a 4- to 12-fold increased risk in VLBW infants (<1,500 g), with odds ratios escalating as birth weight decreases: 4.7 for low birth weight (1,500 to 2,499 g), 11.1 for VLBW (1,000 to 1,499 g), and 12.5 for extremely low birth weight (<1,000 g) infants. Kim et al. [15] suggested that many CH cases in preterm infants were likely transient, as nearly half of treated infants had normal initial thyroid function; however, the permanency of CH was not reported. The notably high incidence of CH observed in our study remains unclear, but it may be partially attributable to the relatively lower gestational age and birth weight of the study cohort. Infants included had a mean gestational age of 28.1 weeks and a mean birth weight of 840 g, placing the majority within the extremely low birth weight infant category. It is well-established that the risk of CH increases with decreasing gestational age and birth weight, further supporting our inference [16]. This elevated risk is driven by multifactorial mechanisms, including HPT axis immaturity, iodine imbalance (deficiency from parenteral nutrition or excess from disinfectants), and comorbidities like RDS [17]. While transient hypothyroidism accounts for most cases in VLBW infants (up to 54.5% in some cohorts), the prevalence of PCH remains debated [18]. Transient cases are linked to delayed TSH elevation, often detected only through repeat screening, whereas PCH in VLBW infants is less common but associated with genetic mutations (e.g., dual oxidase 2 [DUOX2]) or structural thyroid defects. In addition, factors such as iodine excess from betadine use and breast milk feeding may also have contributed to the elevated incidence.

Contrary to our expectation, the mean gestational age in the PCH group was significantly higher than in the TCH group. Furthermore, the presence of major anomalies was associated with PCH in VLBW infants. Previous studies have well documented the association between congenital heart defects and CH [19,20]. In our cohort, among the nine infants with major anomalies, four had congenital heart defects. These findings align with the understanding that PCH is primarily attributed to structural or genetic etiologies. Though the LT4 dose at 36 months was higher in the PCH group in multivariate analysis, this value may assist in deciding whether to perform a trial-off and estimating the likelihood of successful discontinuation, rather than predicting PCH.

The reason for the difference in dexamethasone use between the two groups is unclear. The PCH group had higher gestational age and birth weight compared to the TCH group. Therefore, it is presumed that the risk of developing BPD was relatively lower in the PCH group, which may have led to less frequent use of dexamethasone. Moreover, dexamethasone is known to suppress the HPT axis, which can transiently affect thyroid function and potentially influence classification into the TCH group. However, since the multivariable analysis showed no significant difference in the frequency of dexamethasone use between the two groups, the clinical significance of this finding remains unclear.

The proportion of infants with developmental delay did not differ significantly between the PCH and TCH groups, which may be attributed to timely LT4 supplementation, thereby mitigating potential neurodevelopmental discrepancies. Early treatment of CH has been shown to prevent subsequent NDI [21,22].

To reduce the risk of NDI, many centers continue to maintain LT4 therapy until 36 months of age, which may lead to overtreatment in infants with TCH. Previous studies have also suggested that the longitudinal LT4 dosage requirement may help predict PCH, and some have proposed an earlier trial-off therapy at or before 24 months of age [23-25]. In our study, both an initial screening TSH >10 μIU/mL and higher LT4 dose at 36 months were significantly associated with PCH, effectively distinguishing the PCH group from the TCH group and serving as independent predictors. Our findings may offer a practical framework for developing individualized follow-up strategies in infants with CH (e.g., consideration of early trial-off therapy in those with low screening TSH levels or reduced LT4 requirements at 12 months, versus continuation of therapy with close monitoring in those requiring persistently higher LT4 doses).

This study has some limitations. First, there is a potential for selection bias. As infants who died or were lost to follow-up before reaching 3 years of age were excluded from the analysis, which may have influenced the results. While deaths may reflect underlying clinical severity, most cases of follow-up loss were due to non-clinical factors such as relocation or parental refusal. Moreover, the permanency of thyroid dysfunction (i.e., PCH vs. TCH) is unlikely to be directly associated with mortality or follow-up loss, suggesting that a portion of the missing data may be missing at random rather than entirely systematic, although the possibility of bias cannot be entirely excluded. Another concern for selection bias is including genetic diseases such as trisomy 21 and congenital anomalies. Whether to exclude patients with genetic disorders or major congenital anomalies in epidemiologic studies of CH depends on the study objective. However, such exclusions may lead to underestimation of the actual prevalence and introduce bias, as CH is frequently associated with congenital anomalies. The primary aim of our study was to reflect the actual prevalence of CH in real-world clinical practice, particularly among VLBW infants in whom the rates of genetic and structural anomalies are higher than in term infants [26]. Therefore, we did not exclude patients with genetic syndromes or congenital anomalies. Second, although LT4 dosage at 12 months was examined as a potential predictor for CH permanency, the study did not establish or validate an optimal cutoff value for guiding early treatment withdrawal, indicating a need for further prospective studies. Third, genetic or molecular testing was not performed, which could have contributed to a more precise differentiation of the underlying causes of PCH. Lastly, iodine-related data, such as urinary iodine concentration, the frequency of exposure to povidone-iodine and maternal seaweed soup intake, were not assessed.

In conclusion, the prevalence of PCH in VLBW infants of this study was 1.8%, which is notably higher than previously reported. Higher gestational age, the presence of major congenital anomalies, screening TSH >10 μIU/mL, and higher LT4 requirement dose at 36 months were associated with PCH in VLBW infants. Further large-scale studies are needed to differentiate TCH and PCH, and explore the feasibility of early LT4 trial-off therapy.